# Copyright (c) 2020 zfit import contextlib import itertools import warnings from collections import OrderedDict from typing import Dict, Union, Callable, Optional, Tuple, Iterable import colored import iminuit import numpy as np from colorama import Style, init from ordered_set import OrderedSet from tabulate import tabulate from .errors import compute_errors, covariance_with_weights, dict_to_matrix, matrix_to_dict from .interface import ZfitMinimizer, ZfitResult from ..core.interfaces import ZfitLoss, ZfitParameter from ..core.parameter import set_values from ..settings import run from ..util.container import convert_to_container from ..util.warnings import ExperimentalFeatureWarning from ..util.ztyping import ParamsTypeOpt init(autoreset=True) def _minos_minuit(result, params, sigma=1.0): fitresult = result minimizer = fitresult.minimizer from zfit.minimizers.minimizer_minuit import Minuit if not isinstance(minimizer, Minuit): raise TypeError("Cannot perform error calculation 'minos_minuit' with a different minimizer than" "`Minuit`.") result = [minimizer._minuit_minimizer.minos(var=p.name, sigma=sigma) for p in params][-1] # returns every var result = OrderedDict((p, result[p.name]) for p in params) new_result = None return result, new_result def _covariance_minuit(result, params): if any(data.weights is not None for data in result.loss.data): warnings.warn("The computation of the covariance matrix with weights is still experimental.", ExperimentalFeatureWarning) fitresult = result minimizer = fitresult.minimizer from zfit.minimizers.minimizer_minuit import Minuit if not isinstance(minimizer, Minuit): raise TypeError("Cannot compute the covariance matrix with 'covariance_minuit' with a different" " minimizer than `Minuit`.") covariance = result.minimizer._minuit_minimizer.covariance covariance_dict = {} for p1 in params: for p2 in params: key = (p1, p2) covariance_dict[key] = covariance[tuple(k.name for k in key)] return covariance_dict def _covariance_np(result, params): if any(data.weights is not None for data in result.loss.data): warnings.warn("The computation of the covariance matrix with weights is still experimental.", ExperimentalFeatureWarning) # TODO: maybe activate again? currently fails due to numerical problems # numgrad_was_none = settings.options.numerical_grad is None # if numgrad_was_none: # settings.options.numerical_grad = True _, gradient, hessian = result.loss.value_gradients_hessian(params) covariance = np.linalg.inv(hessian) # if numgrad_was_none: # settings.options.numerical_grad = None return matrix_to_dict(params, covariance) [docs]class FitResult(ZfitResult): _default_hesse = "hesse_np" _hesse_methods = {"minuit_hesse": _covariance_minuit, "hesse_np": _covariance_np} _default_error = "zfit_error" _error_methods = {"minuit_minos": _minos_minuit, "zfit_error": compute_errors} def __init__(self, params: Dict[ZfitParameter, float], edm: float, fmin: float, status: int, converged: bool, info: dict, loss: ZfitLoss, minimizer: "ZfitMinimizer"): """Create a `FitResult` from a minimization. Store parameter values, minimization infos and calculate errors. Any errors calculated are saved under `self.params` dictionary with:: {parameter: {error_name1: {'low': value, 'high': value or similar}} Args: params: Result of the fit where each :py:class:`~zfit.Parameter` key has the value from the minimum found by the minimizer. edm: The estimated distance to minimum, estimated by the minimizer (if available) fmin: The minimum of the function found by the minimizer status: A status code (if available) converged: Whether the fit has successfully converged or not. info: Additional information (if available) like *number of function calls* and the original minimizer return message. loss: The loss function that was minimized. Contains also the pdf, data etc. minimizer: Minimizer that was used to obtain this `FitResult` and will be used to calculate certain errors. If the minimizer is state-based (like "iminuit"), then this is a copy and the state of other `FitResults` or of the *actual* minimizer that performed the minimization won't be altered. """ super().__init__() self._status = status self._converged = converged self._params = self._input_convert_params(params) self._params_at_limit = any(param.at_limit for param in self.params) self._edm = edm self._fmin = fmin self._info = info self._loss = loss self._minimizer = minimizer self._valid = True self._covariance_dict = {} def _input_convert_params(self, params): return ParamHolder((p, {"value": v}) for p, v in params.items()) def _get_uncached_params(self, params, method_name): return [p for p in params if self.params[p].get(method_name) is None] [docs] @classmethod def from_minuit(cls, loss: ZfitLoss, params: Iterable[ZfitParameter], result: iminuit.util.MigradResult, minimizer: Union[ZfitMinimizer, iminuit.Minuit]) -> 'FitResult': """Create a `FitResult` from a :py:class:~`iminuit.util.MigradResult` returned by :py:meth:`iminuit.Minuit.migrad` and a iminuit :py:class:~`iminuit.Minuit` instance with the corresponding zfit objects. Args: loss: zfit Loss that was minimized. params: Iterable of the zfit parameters that were floating during the minimization. result: Return value of the iminuit migrad command. minimizer: Instance of the iminuit Minuit that was used to minimize the loss. Returns: A `FitResult` as if zfit Minuit was used. """ from .minimizer_minuit import Minuit if not isinstance(minimizer, Minuit): if isinstance(minimizer, iminuit.Minuit): minimizer_new = Minuit() minimizer_new._minuit_minimizer = minimizer minimizer = minimizer_new else: raise ValueError(f"Minimizer {minimizer} not supported. Use `Minuit` from zfit or from iminuit.") params_result = [p_dict for p_dict in result[1]] result_vals = [res["value"] for res in params_result] set_values(params, values=result_vals) info = {'n_eval': result[0]['nfcn'], 'n_iter': result[0]['ncalls'], # 'grad': result['jac'], # 'message': result['message'], 'original': result[0]} edm = result[0]['edm'] fmin = result[0]['fval'] status = -999 converged = result[0]['is_valid'] params = OrderedDict((p, res['value']) for p, res in zip(params, params_result)) return cls(params=params, edm=edm, fmin=fmin, info=info, loss=loss, status=status, converged=converged, minimizer=minimizer) @property def params(self): return self._params @property def edm(self): """The estimated distance to the minimum. Returns: Numeric """ return self._edm @property def minimizer(self): return self._minimizer @property def loss(self) -> ZfitLoss: # TODO(Mayou36): this is currently a reference, should be a copy of the loss? return self._loss @property def fmin(self): """Function value at the minimum. Returns: Numeric """ return self._fmin @property def status(self): return self._status @property def info(self): return self._info @property def converged(self): return self._converged @property def valid(self): return self._valid and not self.params_at_limit and self.converged @property def params_at_limit(self): return self._params_at_limit @contextlib.contextmanager def _input_check_reset_params(self, params): params = self._input_check_params(params=params) old_values = run(params) try: yield params except Exception as error: warnings.warn("Exception occurred, parameter values are not reset and in an arbitrary, last" " used state. If this happens during normal operation, make sure you reset the values.", RuntimeWarning) raise set_values(params=params, values=old_values) def _input_check_params(self, params): if params is not None: params = convert_to_container(params) else: params = list(self.params.keys()) return params [docs] def hesse(self, params: ParamsTypeOpt = None, method: Union[str, Callable] = None, error_name: Optional[str] = None) -> OrderedDict: """Calculate for `params` the symmetric error using the Hessian/covariance matrix. Args: params: The parameters to calculate the Hessian symmetric error. If None, use all parameters. method: the method to calculate the covariance matrix. Can be {'minuit_hesse', 'hesse_np'} or a callable. error_name: The name for the error in the dictionary. Returns: Result of the hessian (symmetric) error as dict with each parameter holding the error dict {'error': sym_error}. So given param_a (from zfit.Parameter(.)) `error_a = result.hesse(params=param_a)[param_a]['error']` error_a is the hessian error. """ if method is None: # LEGACY START method = self._default_hesse from zfit.minimizers.minimizer_minuit import Minuit if isinstance(self.minimizer, Minuit): method = "minuit_hesse" # LEGACY END if error_name is None: if not isinstance(method, str): raise ValueError("Need to specify `error_name` or use a string as `method`") error_name = method with self._input_check_reset_params(params) as params: uncached_params = self._get_uncached_params(params=params, method_name=error_name) if uncached_params: error_dict = self._hesse(params=uncached_params, method=method) self._cache_errors(error_name=error_name, errors=error_dict) errors = OrderedDict((p, self.params[p][error_name]) for p in params) return errors def _cache_errors(self, error_name, errors): for param, errors in errors.items(): self.params[param][error_name] = errors def _hesse(self, params, method): covariance_dict = self.covariance(params, method, as_dict=True) return OrderedDict((p, {"error": covariance_dict[(p, p)] ** 0.5}) for p in params) [docs] def error(self, params: ParamsTypeOpt = None, method: Union[str, Callable] = None, error_name: str = None, sigma: float = 1.0) -> OrderedDict: r""" .. deprecated:: unknown Use :func:`errors` instead. Args: params: The parameters or their names to calculate the errors. If `params` is `None`, use all *floating* parameters. method: The method to use to calculate the errors. Valid choices are {'minuit_minos'} or a Callable. sigma: Errors are calculated with respect to `sigma` std deviations. The definition of 1 sigma depends on the loss function and is defined there. For example, the negative log-likelihood (without the factor of 2) has a correspondents of :math:`\Delta` NLL of 1 corresponds to 1 std deviation. error_name: The name for the error in the dictionary. Returns: A `OrderedDict` containing as keys the parameter names and as value a `dict` which contains (next to probably more things) two keys 'lower' and 'upper', holding the calculated errors. Example: result['par1']['upper'] -> the asymmetric upper error of 'par1' """ warnings.warn("`error` is deprecated, use `errors` instead. This will return not only the errors but also " "(a possible) new FitResult if a minimum was found. So change" "errors = result.error()" "to" "errors, new_res = result.errors()", DeprecationWarning) return self.errors(params=params, method=method, error_name=error_name, sigma=sigma)[0] [docs] def errors(self, params: ParamsTypeOpt = None, method: Union[str, Callable] = None, error_name: str = None, sigma: float = 1.0) -> Tuple[OrderedDict, Union[None, 'FitResult']]: r"""Calculate and set for `params` the asymmetric error using the set error method. Args: params: The parameters or their names to calculate the errors. If `params` is `None`, use all *floating* parameters. method: The method to use to calculate the errors. Valid choices are {'minuit_minos'} or a Callable. sigma: Errors are calculated with respect to `sigma` std deviations. The definition of 1 sigma depends on the loss function and is defined there. For example, the negative log-likelihood (without the factor of 2) has a correspondents of :math:`\Delta` NLL of 1 corresponds to 1 std deviation. error_name: The name for the error in the dictionary. Returns: A `OrderedDict` containing as keys the parameter and as value a `dict` which contains (next to probably more things) two keys 'lower' and 'upper', holding the calculated errors. Example: result[par1]['upper'] -> the asymmetric upper error of 'par1' """ if method is None: # TODO: legacy, remove 0.6 from zfit.minimize import Minuit if isinstance(self.minimizer, Minuit): method = 'minuit_minos' warnings.warn("'minuit_minos' will be changed as the default errors method to a custom implementation" "with the same functionality. If you want to make sure that 'minuit_minos' will be used " "in the future, add it explicitly as in `errors(method='minuit_minos')`", FutureWarning) else: method = self._default_error if error_name is None: if not isinstance(method, str): raise ValueError("Need to specify `error_name` or use a string as `method`") error_name = method if method == 'zfit_error': warnings.warn("'zfit_error' is still experimental and may fails.", ExperimentalFeatureWarning) with self._input_check_reset_params(params) as params: uncached_params = self._get_uncached_params(params=params, method_name=error_name) new_result = None if uncached_params: error_dict, new_result = self._error(params=uncached_params, method=method, sigma=sigma) if new_result is None: self._cache_errors(error_name=error_name, errors=error_dict) else: msg = "Invalid, a new minimum was found." self._cache_errors(error_name=error_name, errors={p: msg for p in params}) self._valid = False new_result._cache_errors(error_name=error_name, errors=error_dict) all_errors = OrderedDict((p, self.params[p][error_name]) for p in params) return all_errors, new_result def _error(self, params, method, sigma): if not callable(method): try: method = self._error_methods[method] except KeyError: raise KeyError("The following method is not a valid, implemented method: {}".format(method)) return method(result=self, params=params, sigma=sigma) [docs] def covariance(self, params: ParamsTypeOpt = None, method: Union[str, Callable] = None, as_dict: bool = False): """Calculate the covariance matrix for `params`. Args: params: The parameters to calculate the covariance matrix. If `params` is `None`, use all *floating* parameters. method: The method to use to calculate the covariance matrix. Valid choices are {'minuit_hesse', 'hesse_np'} or a Callable. as_dict: Default `False`. If `True` then returns a dictionnary. Returns: 2D `numpy.array` of shape (N, N); `dict`(param1, param2) -> covariance if `as_dict == True`. """ if method is None: # LEGACY START method = self._default_hesse from zfit.minimizers.minimizer_minuit import Minuit if isinstance(self.minimizer, Minuit): method = "minuit_hesse" # LEGACY END if method not in self._covariance_dict: with self._input_check_reset_params(params) as params: self._covariance_dict[method] = self._covariance(method=method) params = self._input_check_params(params) covariance = {k: self._covariance_dict[method][k] for k in itertools.product(params, params)} if as_dict: return covariance else: return dict_to_matrix(params, covariance) def _covariance(self, method): if not callable(method): try: method = self._hesse_methods[method] except KeyError: raise KeyError("The following method is not a valid, implemented method: {}".format(method)) params = list(self.params.keys()) if any([data.weights is not None for data in self.loss.data]): return covariance_with_weights(method=method, result=self, params=params) else: return method(result=self, params=params) [docs] def correlation(self, params: ParamsTypeOpt = None, method: Union[str, Callable] = None, as_dict: bool = False): """Calculate the correlation matrix for `params`. Args: params: The parameters to calculate the correlation matrix. If `params` is `None`, use all *floating* parameters. method: The method to use to calculate the correlation matrix. Valid choices are {'minuit_hesse', 'hesse_np'} or a Callable. as_dict: Default `False`. If `True` then returns a dictionnary. Returns: 2D `numpy.array` of shape (N, N); `dict`(param1, param2) -> correlation if `as_dict == True`. """ covariance = self.covariance(params=params, method=method, as_dict=False) correlation = covariance_to_correlation(covariance) if as_dict: params = self._input_check_params(params) return matrix_to_dict(params, correlation) else: return correlation def __str__(self): string = Style.BRIGHT + f'FitResult' + Style.NORMAL + f' of\n{self.loss} \nwith\n{self.minimizer}\n\n' string += tabulate( [[color_on_bool(self.valid), color_on_bool(self.converged, on_true=False), color_on_bool(self.params_at_limit, on_true=colored.bg(9), on_false=False), format_value(self.edm, highprec=False), format_value(self.fmin)]], ['valid', 'converged', 'param at limit', 'edm', 'min value'], tablefmt='fancy_grid', disable_numparse=True) string += '\n\n' + Style.BRIGHT + "Parameters\n" + Style.NORMAL string += str(self.params) return string def _repr_pretty_(self, p, cycle): if cycle: p.text(self.__repr__()) return p.text(self.__str__()) [docs]def covariance_to_correlation(covariance): diag = np.diag(1 / np.diag(covariance) ** 0.5) return np.matmul(diag, np.matmul(covariance, diag)) [docs]def format_value(value, highprec=True): try: import iminuit m_error_class = iminuit.util.MError except ImportError: m_error_class = dict if isinstance(value, (dict, m_error_class)): if 'error' in value: value = value['error'] value = f"{value:> 6.2g}" value = f'+/-{" " * (8 - len(value))}' + value if 'lower' in value and 'upper' in value: lower = value['lower'] upper = value['upper'] lower_sign = f"{np.sign(lower): >+}"[0] upper_sign = f"{np.sign(upper): >+}"[0] lower, upper = f"{np.abs(lower): >6.2g}", f"{upper: >6.2g}" lower = lower_sign + " " * (7 - len(lower)) + lower upper = upper_sign + " " * (7 - len(upper)) + upper # lower += " t" * (11 - len(lower)) value = lower + " " * 3 + upper if isinstance(value, float): if highprec: value = f"{value:> 6.4g}" else: value = f"{value:> 6.2g}" return value [docs]def color_on_bool(value, on_true=colored.bg(10), on_false=colored.bg(9)): if not value and on_false: value_add = on_false elif value and on_true: value_add = on_true else: value_add = '' value = value_add + str(value) + Style.RESET_ALL return value [docs]class ParamHolder(dict): # no UserDict, we only want to change the __str__ def __str__(self) -> str: order_keys = ['value', 'hesse'] keys = OrderedSet() for pdict in self.values(): keys.update(OrderedSet(pdict)) order_keys = OrderedSet([key for key in order_keys if key in keys]) order_keys.update(keys) rows = [] for param, pdict in self.items(): row = [param.name] row.extend(format_value(pdict.get(key, ' ')) for key in order_keys) row.append(color_on_bool(run(param.at_limit), on_true=colored.bg('light_red'), on_false=False)) rows.append(row) order_keys = ['name'] + list(order_keys) + ['at limit'] table = tabulate(rows, order_keys, numalign="right", stralign='right', colalign=('left',)) return table